Better persistence

[jgraeger]

After merging #9, the service will persist data to disk. However, the implementation is very naive. Read times will decrease over time, as for each database query, the service will read through the whole file. Second, everytime a whole new record containing each key for each known service will created. This isn`t a problem per sé und maybe we should leave it like that for simplicity. However, the database file size will grow and grow over time. The persistence layer needs to be improved.

We want to:

• Significantly improve read performance (most obvious solution: Create a database index)
• Squash records that are not needed anymore, but keep them accessible in some way. Using an append-only DB is intended.